Beta

How to get started with ad testing to create more effective ads

Multivariate testing measures the performance of every possible combination of creative elements. Here’s how to get started testing and optimizing your own ads.
Susan Wenograd

Ads are one of the most important tools a website or business can use to increase traffic and conversions, and creating ads that resonate with your target audience is essential for a successful ad campaign. But it's not easy — that's where ad testing comes in. In this post, we'll outline the different types of ad testing, explain the benefits of each, and offer some tips to get started.

What is multivariate ad testing?

We’re going to cut to the chase here: if you want to improve your marketing strategy in a systematic way, then multivariate testing is the only way to go.

We recommend using multivariate ad testing because it allows you to test all possible combinations of your ad creative at the same time. You get very specific insights about your ad creative that can then be applied to the design of future ads. 

Essentially, this process allows you to find out not only which of your ads are winners, but also why they are winners — was it the headline, the image, the color? With this information, you can be sure you are running the most effective ads possible. 

Multivariate testing vs. A/B testing

Multivariate testing and A/B testing are often mistakenly seen as the same type of ad testing, but there is a key difference between the two. 

A/B testing is the testing of only two different ads against each other, therefore it can only provide you with a very limited amount of information — basically, which one is better than the other. Most importantly, it cannot tell you what specific elements in the winning ad are responsible for its success.

Multivariate testing, on the other hand, can give you that information, as it compares all the possible options against each other, so more specific data that can help you create more winning ads. Consequently, multivariate testing provides better results than A/B testing if done correctly. 

Multivariate testing vs. dynamic creative optimization (DCO)

Multivariate testing and dynamic creative optimization (DCO) are also often mistakenly seen as the same thing, but they actually perform two different tasks. 

Multivariate testing is a method of testing multiple ad variations to see which one produces the best results.

DCO uses artificial intelligence (AI) to adapt the ad shown to a user in real time, based on contextually relevant factors such as location or the weather. 

These options aren’t mutually exclusive – both methods actually work quite well together. You can use multivariate testing to determine which ad elements make a high-performing ad, and then use DCO to ensure that ad will be shown to the right people, at the right time.

For example, maybe you use a tool like Marpipe. You run a multivariate test with 3 different CTAs, 3 different images, and 3 different fonts.

After completing that test, you know which of those 27 ad variants (3x3x3) are the best for that audience. 

At that point, you would winning ads to feed into your DCO. Then, the DCO is going to determine which of those 4 ads is best to show based on the person’s location.

Basically, DCO is good to determine when or where to show the winning ad, BUT multivariate testing is what helps you find and create that winning ad in the first place.

What is tested in a multivariate test?

Now that you know what a multivariate test is, let's take a closer look at some elements involved in one.

In every multivariate test, you have your independent variables, dependent variables, and fixed variables. 

An independent variable is something that changes in each test (for example, the image used), while the dependent variable is the measured result of changing that independent variable (for example, the click-through-rate). The fixed variable is the parts of your ad that remain the same throughout all of the tests.

That being said, here are some examples of variables in a multivariate test:

Examples of independent variables that are often tested (changed) would be:

  • The graphic or image used
  • The CTA
  • Color of the button or text
  • Font used

Examples of common dependent (measured) variables:

  • Click-through-rate (CTR)
  • Conversion Rate
  • Engagement
  • Cost-per-click (CPC)
  • Cost-per-acquisition (CPA)

In the simplest terms, the independent variable(s) is what you are manipulating or changing, and the dependent variable is what is being measured. The fixed variables aren't really ‘variables’ at all, because they will stay the same across all your ads throughout the entire test.

The role of templates in a multivariate ad test

Since multivariate tests involve testing multiple iterations of the same ad, be sure you are using elements that can be easily interchanged with a still-aesthetically pleasing end result, and all the elements make sense together. The best time-saving hack for this part of the process is to use a design template. 

By using pre-made design templates, you can reduce the time needed to set up and design all of your ad variants. Your template design needs to be well-thought out, because it needs to be able to allow for the creative elements to be interchanged. With a poorly designed template, you could end up with ads that have different elements which don’t fit into the template properly, or elements that don’t make sense together. This results in running an ad that looks like it was pieced together, and isn’t going to give you great results. Getting the template right is essential for getting accurate insights from your ad tests.

Before you start testing, set objectives

Before you start testing, spend some time setting your goals and objectives for your advertising. 

Setting objectives is essential to the success of your testing campaign. Otherwise, it can become difficult to determine whether or not your ads are achieving their desired results. Without a specific goal or objective for your ads, it makes it hard to put together effective and efficient marketing campaigns. By taking the time to set your objectives, you’ll have a better idea of where you're headed when it comes time to start testing.

Objectives should be clear, measurable, and attainable. Some common goals for advertising include:

  • Increasing leads
  • Driving higher sales
  • Improving advertising ROI 
  • Increasing traffic to website

The objective that you choose all depends on the stage your business is at, and what you are looking to prioritize. 

For example, for early-stage startups, it may be that the most important thing is to improve the awareness of your company, or to have more traffic coming to your website. For companies that aren’t as new, they might be looking to just improve their advertising ROI, since their brand is already pretty established in the market. There isn’t really a right or wrong objective, but whatever you decide, it should be aligned with helping you meet your overall business goals. 

Which ads should you test first?

It can be tough to know where to start when it comes to ad testing, but in general, the most effective method is to start out with your top-performing ads first. Hey, if it ain’t broke, don’t fix it, right?

Tips for your first ad test

Hypothesis

Before starting your ad testing, first come up with a hypothesis. This will help you understand the results of your tests and help you make more informed decisions about which ads work best. A good hypothesis should be based on logical inferences made from past observations of ad performance — they shouldn't just be totally random guesses or ideas. Be sure the hypothesis you choose is one that is actually testable, and one that is relevant to your business. 

Examples of strong hypothesis are:

  • Testing specific headlines for an increase in conversion rate
  • Testing button size or color increases the click-through rate

Taking time to formulate a well-thought out hypothesis will help you keep you focused and your ad testing on track. 

Duration

A common question many people have when they start testing ads is: how long should I run the test for? 

While there isn’t any real hard and fast rule, we usually recommend running your first test for 7 days to get a baseline measure of creative intelligence (CQ). After that, you can analyze the results and adjust your strategy or test duration accordingly.. 

Factors to consider when planning your ad test

Every brand has different logistical requirements to consider and take into account when you are planning the test. 

Somethings to consider when planning your test include: your ad testing budget, the audience you target, the number of variants, and the amount of time you run the test. All of these elements affect each other - you can’t change one of them without affecting the rest. This is why it is important to spend the time planning your test out before you start – figuring it out on the fly leads to some not-so-great results. 

Budget

Your budget helps you determine two things: the length of your test, and also your budget per ad group.

There are a few things to know about Marpipe that will help you with setting an adequate budget: 

  • Marpipe can run 7- or 14-day ad testsYou should budget at least 1x your average CPA per ad variant, per test. This is because it is the minimum that will give you a baseline of creative intelligence.
  • We recommend a budget 2x your average CPA per ad variant, per test, so you can collect more valuable data (this ensures each ad variant will have at least two conversions)
  • The higher your budget is per test, the more variables you can have per test. 

If setting your ad test budget according to your average CPA seems confusing, don’t worry. Marpipe will show you how your budget actually breaks down before you start the test, so you can always alter things if need be. 

Audience

For your first test, run your ads in front of an audience you already know is likely to perform well. This helps you get a solid baseline of helpful data to use when you are planning future tests and experimenting with other variables. 

Remember if you are changing your audience for each ad, your total number of ad variants is going to multiply as well, because the audience is then treated as a creative variant. 

Variant volume

It’s important that you don’t get too carried away when you are choosing your creative variants, and consider how it will affect your budget allocation. 

For example, if you wanted to test out 3 call-to-actions, 3 different headlines, and 3 different images, that means you’d be testing 18 different ads in your test. 

3 call-to-actions x 2 different headlines x 3 different images = 27 ad variants 

And with a budget of $1000, this would allot $37 ad spend for each ad variant. 

However, let’s say you want to add another creative variant to that test, and you want to also test out 4 different background colors. Now suddenly you are looking at testing 108 different ad variants. 

3 call-to-actions x 3 different headlines x 3 different images x 4 different colors = 108 ad variants 

In this scenario, your $1000 wouldn’t go quite as far, leaving you just over $9 per ad. Not ideal. 

So, we usually will recommend that you run a total of 20-30 ads that are made up of 2-3 variant groups with 2-4 variants each. 

Cadence

Cadence is the length of your ad testing time, and as mentioned above, there are two test lengths available when you use Marpipe: 7 days or 14 days. 

Generally, your test length is determined by your ad testing goals as well as the budget you set, but sometimes it makes sense to go with a particular one over another. 

For example, if you have a relatively short time frame to run your ad, you would want to go with a shorter test in order to turn the ad around quickly and scale it.

To see how the two test lengths are used, here’s a handy-dandy chart:

Beginning the Marpipe process

If you’ve read this far, then you are probably already pretty certain about the awesomeness of Marpipe. But in case you forgot:

If you're looking to gain data-driven insights about how your ad creative is performing, then Marpipe is the tool for you. With its multivariate testing method, and ability to track a variety of metrics like click-through rate (CTR), conversion rate, and spending on ads, Marpipe gives you detailed insights that to make informed decisions about your ad creative. 

When you first start testing your ads with Marpipe, you need to do a couple of things before running your ad tests. This includes creating your ad templates, and determining which creative elements you are going to use as variables in your testing. 

Once you have your template set up, and have decided which parts you are going to change in the ads, then you add the specific creative elements into Marpipe. Finally, generate the variants for the test, and you can launch them directly into Facebook and Instagram. 

Start testing and optimizing your ads

Now that you understand multivariate testing and how to use an automated tool like Marpipe during the testing process, you are well on your way to creating more winning ads (and enjoying the aforementioned awesomeness of Marpipe!)

How to Run a Multivariate Test

The Beginner's Guide

Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.
Tiffany Johnson Headshot

How to Run a Multivariate Test
The Beginner's Guide

Plus, Get our Weekly
Experimentation newsletter!

Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.